I'm applying the PageRank algorithm to a Natural Language Processing problem; basically to rank words like "love", "adore" and "like" in terms of their strength or intensity.
I weighted links/edges using Pointwise Mutual Information which gives me a rank correlation of 0.482 using Spearmans Rank Correlation.
A file of node absolute frequencies (counts)
A file of link absolute frequencies (counts)
A file of the human-generated rank (the benchmark)
A code for PageRank computation
A code to evaluate the PR rank using Spearman Rank Correlation
All I need is someone to use PyBrain Machine learning library and particularly its optimization algorithms (already implemented) to find the best way to weight nodes and links because my PMI method gives me only 0.482
No need of natural language processing is required. I can explain it. Experience with Machine learning is a must.
Budget limited to $100